Determining risk appetitezz

A bank must limit the amount it is prepared to lose under extreme market circumstances;
this can be done using stress test limits. As with all limit setting, the process
should start with the identification of the bank’s risk appetite.
If we start with the premise that a bank’s risk appetite is expressed as a monetary
amount, let us say $10 million, then a natural question follows. Are you prepared to
lose $10 million every day, once per month, or how often? The regularity with which
a loss of a given magnitude can be tolerated is the key qualification of risk appetite.
Figure 8.10 shows how a bank’s risk appetite could be defined. Several different
losses are identified, along with the frequency with which each loss can be tolerated.
The amount a bank is prepared to lose on a regular basis is defined as the 95% confidence VaR (with a one-day holding period). The daily VaR limit should be set
after consideration of the bank’s, or trading unit’s profit target. There would be little
point in having a daily VaR limit that is larger than the annual profit target, otherwise
there would be a reasonable probability that the budgeted annual profit would be
lost during the year.

The second figure is the most important in terms of controlling risk. An institution’s
management will find it difficult to discuss the third figure – the extreme tolerance
number. This is because it is unlikely that any of the senior management team have
ever experienced a loss of that magnitude. To facilitate a meaningful discussion the
extreme loss figure to be used to control risk must be of a magnitude that is believed
possible, even if improbable.

The second figure is the amount the bank is prepared to lose on an infrequent
basis, perhaps once every two or three years. This amount should be set after
consideration of the available or allocated capital. This magnitude of loss would arise
from an extreme event in the financial markets and will therefore not be predicted
by VaR. Either historic price changes or extreme value theory could be used to
predict the magnitude. The bank’s actual capital must more than cover this figure.
Stress tests should be used to identify scenarios that would give rise to losses of this
magnitude or more.

Once the scenarios have been identified, management’s subjective judgement must
be used, in conjunction with statistical analysis, to judge the likelihood of such an
event in prevailing market conditions. Those with vested interests should not be
allowed to dominate this process. The judgement will be subjective, as the probabilities
of extreme events are not meaningful over the short horizon associated with
trading decisions. In other words the occurrence of extreme price shocks is so rare
that a consideration of the probability of an extreme event would lead managers to
ignore such events. EVT could be used in conjunction with management’s subjective
judgement of the likelihood of such an event, given present or predicted economic
circumstance.

The loss labelled ‘extreme tolerance’ does not have an associated frequency, as it
is the maximum the bank is prepared to lose – ever. This amount, depending on the
degree of leverage, is likely to be between 10% and 20% of a bank’s equity capital
and perhaps equates to a 1 in 50-year return level. Losses greater than this would
severely impact the bank’s ability to operate effectively. Again stress tests should be
used to identify scenarios and position sizes that would give rise to a loss of this
magnitude. When possible (i.e. for single assets or indices) EVT should then be used
to estimate the probability of such an event.